Tape drive cleaning

[Amaan]

Hi,

I need help for another issue (actually may ot be issue). We have lto5 tape drives (iscalar6000 tape library). drives are realy busy. Issue is with cleaning. we are running cleaning from NetBackup, but it is asking to clean it again after 1 day, 2 days, week. I dont think that it should be so frequently. What can be reason for this?

Please find screenshot of the device monitor:

Thanks in advance!

[Yasuhisa_Ishika]

Check frequency setting for each drive first. Can you post output of commands below?

# tpclean -L
# tpconfig -emm_dev_list

Also, check your library and drives from management tool whether it really requst cleaning.

[Amaan]

Here are the requested outputs:

F:\Veritas\Volmgr\bin>tpclean.exe -L
Drive Name              Type      Mount Time  Frequency   Last Cleaned         Comment
**********              ****      **********  *********   ****************     *******
HP.ULTRIUM5-SCSI.009    hcart2*   374.8       0           17:27 01/03/2012     NEEDS CLEANING
HP.ULTRIUM5-SCSI.008    hcart2*   129.1       0           18:23 01/16/2012     NEEDS CLEANING
HP.ULTRIUM5-SCSI.007    hcart2*   452.3       0           16:51 01/03/2012     NEEDS CLEANING
HP.ULTRIUM5-SCSI.006    hcart2*   457.4       0           16:42 01/03/2012     NEEDS CLEANING
HP.ULTRIUM5-SCSI.002    hcart2*   0.0         0           00:03 01/26/2012
HP.ULTRIUM5-SCSI.004    hcart2*   110.9       0           12:38 01/19/2012
HP.ULTRIUM5-SCSI.003    hcart2*   9.6         0           00:02 01/26/2012     NEEDS CLEANING
HP.ULTRIUM5-SCSI.005    hcart2*   303.2       0           18:42 01/10/2012     NEEDS CLEANING

Second output is attached.

And can you let me know what is management tool. Do you mean library console?

[Yasuhisa_Ishika]

NetBackup 7.1Administrator's Guide for Windows Volume II , page 123
NetBackup reports that a drive needs cleaning if either of the following conditions
are true:
■ The value for the mount time is greater than the cleaning frequency.
■ The TapeAlert CLEAN_NOW or CLEAN_PERIODIC flag is set.
And either of the following conditions must be true:
■ The drive is a standalone drive and a cleaning tape is not defined.
■ The drive is a standalone drive and no cleaning tape has any cleanings that
remain.

Try to set Cleaning Frequency for each drive. Setting too short frequency is harmful, it's better to set it to thousands of hours.
Setting loger frequency is not problem. In situation that cleaning is really required, tape drive request cleaning by tape alert, and request based cleaning will be scheduled while mount time does not reach frequency.

And for now, to clear NEED CLEANING flag, run "tpclean -M <drive_name>" so that it reset mount time.

[Marianne]

I do not agree with Frequency cleaning.

I believe that the default TapeAlert should be used.

Frequent cleaning requests is normally a result of slow throughput.

Some Cleaning TN's:

 http://www.symantec.com/docs/TECH19399

http://www.symantec.com/docs/TECH37409.

See http://www.symantec.com/docs/HOWTO56118 for explanation of cleaning TapeAlert messages.

[Yasuhisa_Ishika]

Hello, Marianne.
I agree with you, short frequency is harmful. But also I think frequency of tens of thousand is substantially equal to disableing frequency base cleaning.
For veterans here, NEEDS CLEANINGS is not matter(judge with mount time and error log). But I would like to tell him how to avoid this flag.
 

[mph999 · Accepted solution]

Amaan,

It is the drive itself that is requesting cleaning - via a Tapealert.

The reason would have to be determined by your hardware vendor - it is their tapedrive, and their tapealert, not NetBackups.

Reasons that are  common,

Drives are wearing out, producing many errors (you won;t see these errors) which are detected by the firmware and therefore it requests cleaning.

Old media, that is leaving a lot of debris (oxcide off the tape) in the drive

Firmware issue

Drives are simply used a lot, and cleaning is required

It is impossible to find a reason in Netbackup.  All we do, is clean the drives, when told to via the Tapealert sent directly from the firmware of the drive.

Martin

[Mark_Solutions]

From your figures this looks like a fairly new tape library to me - mount time of 0 hours on one drive and 9.6 on another, but still only 450 hours at most (19 days constant use)

That being the case it could be one of 3 things:

1. New drives no longer go through a cleaning cycle before being shipped due to the cost - so sacrifice some cleaning tapes and run three cleaning against each drive to see if it resolves it for you (this is what used to happen before manufacturer cut backs)

2. As they are LTO5 and the library is new I am guessing that they are also new tapes - in that case they may not be the best quality and may be making the situation worse - take a look at the RAS tickets on the tape library to see how many tape and drive alerts are present on it and what they are - it is worth paying the extra for really good tapes!

3. You have have a bad batch of tape drives (very common these days unfortunately) - raise a case with Quantum and get them all upgraded to the latest firmware and insist on sending them a library snapshot (you do this via the library web GUI) for examination. They may ask you to run the HP Tape tools on the drives as well - it is worth doing. In one environment I manage we have replaced at least 22 out of 18 HP drives so far (yes - some have been repalced at least twice!) - get a case logged and make the most of your warranty.

Hope this helps

[mph999]

Regarding Marks post above, specifically (2).

What make are the tapes (this can be seen in bptm log at VERBOSE = 5).

I'm going to hazard a guess here ...  that this is perfectly normal ' Drives are real busy ..."  - Well they'll need cleaning then ...  But first, I must ask ...

" I dont think that it should be so frequently ..."

Why not - how long has this system been busy ?  Has the amount of tape cleaning requests increased ?

Cheers,

Martin

[matt077]

customer of ours had the same kind of problem

One drive kept asking for a clean every 1 or 2 days, when we looked into this, it was the tape drive what was at fault, had the drive replacement and problem was fix.

[js88699]

My opinion is if the library and drives can manage their own cleaning, then get NetBackup out of the hardware business.  I've been told by several techs that hardware cleaning is the preferred method.

[mph999]

Not sure why, it shouldn't make a difference.

I agree that it is more simple to let the library do the cleaning, simply because there uis less config involved to set up, but from the technical view point it makes no difference at all.

1.  The drive needs cleaning and send out a tape alert.

2.  Depending on who does the cleaning, this tapealert is detected by either the library, or, NetBackup and the cleaning tape is loaded.

3.  The tapealert is then cleared.

So, just because NetBackup may be doing the cleaning, it does not change how often theh drive needs cleaning. 

How the cleaning is dset up defines  WHO loads the tape, not when the tape is loaded, providing of course, frequency based cleaning is not being used (which is NOT recommened ).

The answer to this post is very simple.

If the drives are being overcleaned, or, you are concerned about how often they are being cleaned, speak with the hardware vendor, as it is the firmware of the tape deives that are sending out the cleaning request.

NetBackup has no control over when the drives need cleaning (again, providing frequency based cleaning is NOT being used).

The solution will be either ...

Faulty drive / hardware issue

Drive is worn out (this can cause 'false' cleaning requests).

Firmware issue

Cleaning frequency is perfently normal as drives are used often

Martin

[Yasuhisa_Ishika]

https://www-secure.symantec.com/connect/forums/could-not-write-tape-mark-begin-new-image Cleaning is really requested by TapeAlert so often. This is not matter of Cleaning Frequency setting. CERTANCE Ultrium 2 - I suspect that you have used this several years, and tape drives or tapes or both have been worn out.

[Amaan]

Tape drives which are mentioned here and in the other thread are all new. they have been used just 3-5 months.

[Marianne]

Even more reason to log a support call with your hardware vendor... it could be a firmware problem.